A comparison of forecasting methods for daily air quality index using artificial neural network and a hybrid of artificial neural network and Markov chain model
Abstract:
The objective of this research was to compare a daily air quality index forecasting model considering the mean absolute percentage error. Two forecasting techniques were studied, namely the artificial neural network and the hybrid of artificial neural network and
markov chain model. Data used in the research study were secondary data such as wind speed, temperature, relative humidity, month and the air pollutant toxic concentrations measured at the station located at the Public Relations Department, Phaya Thai Subdistrict,
Phaya Thai District, Bangkok from January 1, 2019, to November 30, 2020. A total of 700 entries were divided into two sets : training datasets and testing datasets. The results showed that the hybrid of artificial neural network and markov chain model yielded a mean
absolute percentage error of 23.75%, while the artificial neural network model yielded a mean absolute percentage error of 25.51%. It can be said that forecasting the daily air quality index using the hybrid of artificial neural network and markov chain forecast model
is more accurate than the artificial neural network model. The forecast model can be used to forecast the daily air quality index in advance.